papers checked against Recognition

60 of 33987 papers

• Quantum Thermal State Preparation

  quant-ph · 2023-03-31 · 3 theorem links

  Continuous-time quantum Gibbs samplers prepare thermal states and purified thermal field double states with provable costs depending on temperature, accuracy and mixing time, plus the first rigorous finite-time thermalization proof for physically derived Lindbladians.

• Quantum harmonic oscillator, index theorem and spectral asymmetry

  quant-ph · 2025-05-27 · 3 theorem links

  Partition function of the quantum harmonic oscillator is identified as the Chern character of a virtual physical sheaf, establishing a link to the index theorem.

• Uncertainty Estimation for Heterophilic Graphs Through the Lens of Information Theory

  cs.LG · 2025-05-28 · 3 theorem links

  A post-hoc density estimator on the joint space of all MPNN layer embeddings delivers state-of-the-art epistemic uncertainty on heterophilic graphs while matching prior performance on homophilic graphs.

• On the geometry of synthetic null hypersurfaces

  math.DG · 2025-06-05 · 3 theorem links

  Introduces a synthetic null energy condition using optimal transport on topological causal spaces that agrees with the classical NEC in smooth cases and enables proofs of area and singularity theorems in non-smooth settings.

• WISCA: A Consensus-Based Approach to Harmonizing Interpretability in Tabular Datasets

  cs.LG · 2025-06-06 · 3 theorem links

  WISCA produces consensus explanations on synthetic tabular datasets that align with the most reliable individual interpretability method among those tested.

• DVP-MVS++: Synergize Depth-Normal-Edge and Harmonized Visibility Prior for Multi-View Stereo

  cs.CV · 2025-06-16 · 3 theorem links

  DVP-MVS++ combines depth-normal-edge alignment via erosion-dilation and harmonized visibility priors with geometry consistency checks to achieve state-of-the-art multi-view stereo results on ETH3D, Tanks & Temples, and Strecha.

• ViTacFormer: Learning Cross-Modal Representation for Visuo-Tactile Dexterous Manipulation

  cs.RO · 2025-06-19 · 3 theorem links

  ViTacFormer learns a cross-modal visuo-tactile latent space with autoregressive tactile prediction and an easy-to-hard curriculum, then uses the representation for imitation learning that yields ~50% higher success and the first reported 11-stage, 2.5-minute autonomous dexterous tasks.

• Fully convolutional 3D neural network decoders for surface codes with syndrome circuit noise

  quant-ph · 2025-06-19 · 3 theorem links

  A 3D convolutional neural network decoder for surface codes with circuit noise generalizes to distance-97 codes with thresholds up to 0.7% depolarizing noise and improved latency over MWPM above distance 33.

• Brown-Voiculescu entropy revisited

  math.OA · 2025-06-23 · 3 theorem links

  Introduces coloured noncommutative entropies and proves variational principles plus generic infinite entropy for endomorphisms of certain classifiable C*-algebras.

• Step-Video-T2V Technical Report: The Practice, Challenges, and Future of Video Foundation Model

  cs.CV · 2025-02-14 · 3 theorem links

  Step-Video-T2V describes a 30B-parameter text-to-video model with custom Video-VAE, 3D DiT, flow matching, and Video-DPO that claims state-of-the-art results on a new internal benchmark.

• LongWriter-Zero: Mastering Ultra-Long Text Generation via Reinforcement Learning

  cs.CL · 2025-06-23 · 3 theorem links

  LongWriter-Zero applies RL from a base model with specialized rewards for length, quality, and structure to outperform SFT baselines and larger models on long-writing benchmarks.

• Microscopic origin of the magnetic interactions and their experimental signatures in altermagnetic La$_2$O$_3$Mn$_2$Se$_2$

  cond-mat.str-el · 2025-06-26 · 3 theorem links

  Ab initio plus analytical analysis shows multiorbital Mn2+ character produces dominant antiferromagnetic nearest-neighbor exchange in La2O3Mn2Se2 altermagnet, yielding identifiable magnon signatures.

• Logit-Gap Steering: A Forward-Pass Diagnostic for Alignment Robustness

  cs.CR · 2025-06-30 · 3 theorem links

  The refusal-affirmation logit gap quantifies alignment safety margins on toxic prompts, and logit-gap steering efficiently discovers transferable in-distribution suffixes that close the gap and yield 38-96% True ASR with low perplexity.

• R1-Zero's "Aha Moment" in Visual Reasoning on a 2B Non-SFT Model

  cs.AI · 2025-03-07 · 3 theorem links

  RL on Qwen2-VL-2B with SAT dataset produces R1-like reasoning and 59.47% CVBench accuracy, outperforming base model by ~30% and SFT by ~2%.

• Unveiling the Gravitational Universe at \mu-Hz Frequencies

  astro-ph.IM · 2019-08-29 · 3 theorem links

  Proposal for a μ-Hz space-based gravitational wave interferometer to observe massive black hole binaries in early inspiral and low-frequency galactic binaries.

• Variational subspace methods and application to improving variational Monte Carlo dynamics

  quant-ph · 2025-07-11 · 3 theorem links

  A direct subspace optimization formalism is developed and applied via the Bridge method to mitigate discretization errors in variational Monte Carlo dynamics.

• Time Series Forecasting Through the Lens of Dynamics

  cs.LG · 2025-07-21 · 3 theorem links

  Proposes dynamics-based analysis of time series models showing partial dynamics learning and end-positioning as key to performance, plus a plug-and-play improvement method.

• Foundations of Relational Quantum Field Theory I: Scalars

  quant-ph · 2025-07-29 · 3 theorem links

  A relational quantum field theory for scalars is built from Poincaré-covariant quantum reference frames, yielding local observables and fields that satisfy causality and reproduce key Wightman and Algebraic QFT properties.

• A Bayesian approach to model uncertainty in single-cell genomic data

  q-bio.GN · 2025-08-04 · 3 theorem links

  A variational Bayesian framework using Gaussian mixture models for probabilistic clustering of single-cell RNA-seq data, applied to neurogenesis and breast cancer with proposed misclustering rate and AUC metrics.

• An analytical framework for the Levine hats problem: new strategies, bounds and generalizations

  math.CO · 2025-08-03 · 3 theorem links

  Develops an integral framework for Levine hat strategies, constructs recursive S_5 achieving 7/20 for two players with improved convergence, and proves optimal probability exactly 1/2 for uncountable stacks.

• SkySplat: Generalizable 3D Gaussian Splatting from Multi-Temporal Sparse Satellite Images

  cs.CV · 2025-08-13 · 3 theorem links

  SkySplat integrates the RPC camera model into a generalizable 3D Gaussian Splatting pipeline with cross-self consistency masking and multi-view aggregation to reconstruct 3D scenes from sparse satellite images using only RGB and relative height cues.

• Unified theory of classical and quantum ergotropy

  cond-mat.stat-mech · 2025-08-18 · 3 theorem links

  A general analytical expression for classical ergotropy is derived and shown to be the classical limit of the quantum expression for ergodic systems, with the coherent-incoherent decomposition persisting classically.

• EventTSF: Event-Aware Non-Stationary Time Series Forecasting

  cs.LG · 2025-08-19 · 3 theorem links

  EventTSF combines time series and textual events in a diffusion framework with event-conditioned timestep sampling, reporting 41.3% average gains in probabilistic forecasting over 12 baselines on 7 datasets.

• Smooth Flow Matching for Synthesizing Functional Data

  stat.ML · 2025-08-19 · 3 theorem links

  SFM generates smooth functional data via a parsimonious copula-based flow that avoids Gaussianity and low-rank assumptions while handling irregular sampling efficiently.

• ProRL: Prolonged Reinforcement Learning Expands Reasoning Boundaries in Large Language Models

  cs.CL · 2025-05-30 · 3 theorem links

  Prolonged RL training with KL control and reference policy resetting enables LLMs to develop novel reasoning strategies inaccessible to base models even under extensive sampling.

• The Page curve of Hawking radiation from semiclassical geometry

  hep-th · 2019-08-29 · 3 theorem links

  Hawking radiation entropy follows the Page curve when quantum extremal surfaces are identified with RT/HRT surfaces in a higher-dimensional holographic dual, making the black hole interior part of the radiation's entanglement wedge.

• Towards a Unified Framework for Pseudo-Nambu-Goldstone Dark Matter and Electroweak Baryogenesis

  hep-ph · 2025-09-01 · 3 theorem links

  The cS2HDM unifies a pseudo-Nambu-Goldstone dark matter candidate with electroweak baryogenesis in a two-Higgs-doublet plus complex singlet setup, featuring naturally suppressed DM-nucleon scattering and CP-violating Higgs interactions under flavour alignment.

• Improving Generative Methods for Causal Evaluation via Simulation-Based Inference

  cs.LG · 2025-09-02 · 3 theorem links

  SBICE uses simulation-based inference to infer posteriors over generative methods and parameters, producing synthetic causal datasets whose estimates align closely with source data.

• Tuning Nonradiative Recombination via Cation Substitution in Inorganic Antiperovskite Nitrides

  cond-mat.mtrl-sci · 2025-09-04 · 3 theorem links

  Cation substitution in X3NSb antiperovskites (X=Ca, Sr, Ba) tunes nonradiative recombination lifetimes via changes in band gap, NA coupling, and decoherence, with hexagonal Sr3NSb showing the longest lifetime.

• Quasicoherent states of noncommutative D2-branes, Aharonov-Bohm effect and quantum Mobius strip

  math-ph · 2025-09-09 · 3 theorem links

  An analytical formula for quasicoherent states on noncommutative D2-branes is derived and used to exhibit Aharonov-Bohm-type effects on cylinder and Mobius strip geometries.

• A Path Signature Framework for Detecting Creative Fatigue in Digital Advertising

  stat.AP · 2025-09-11 · 3 theorem links

  The paper introduces a path signature-based method for detecting creative fatigue in digital ads using geometric change detection on performance paths.

• SWE-QA: Can Language Models Answer Repository-level Code Questions?

  cs.CL · 2025-09-18 · 3 theorem links

  SWE-QA creates a new repository-level code QA benchmark with 576 pairs and an agentic LLM framework, showing promise but open challenges for models handling complex codebases.

• Cross-Distribution Diffusion Priors-Driven Iterative Reconstruction for Sparse-View CT

  eess.IV · 2025-09-16 · 3 theorem links

  CDPIR integrates cross-distribution diffusion priors from a Scalable Interpolant Transformer trained with classifier-free guidance into model-based iterative reconstruction to improve sparse-view CT under out-of-distribution conditions.

• Similarity Field Theory: A Mathematical Framework for Intelligence

  cs.AI · 2025-09-21 · 3 theorem links

  Similarity Field Theory defines a similarity field over entities, concepts as superlevel-set fibers, and intelligence as a generative operator that preserves fiber membership under evolution.

• Incomplete Data, Complete Dynamics: A Diffusion Approach

  cs.LG · 2025-09-24 · 3 theorem links

  A conditional diffusion model trained on partitioned incomplete samples for physical dynamics achieves asymptotic convergence to the true generative process under mild conditions and outperforms baselines in imputation.

• ff-bifbox: A scalable, open-source toolbox for bifurcation analysis of nonlinear PDEs

  math.NA · 2025-09-22 · 3 theorem links

  ff-bifbox is a new open-source toolbox that enables scalable bifurcation and stability analysis of large time-dependent nonlinear PDEs discretized with adaptive finite elements.

• AgenticEval: Toward Agentic and Self-Evolving Safety Evaluation of Large Language Models

  cs.AI · 2025-09-30 · 3 theorem links

  AgenticEval is a multi-agent framework that ingests unstructured policies to generate and self-evolve comprehensive safety benchmarks for LLMs, with experiments showing declining safety rates as tests harden.

• Minimal model of self-organized clusters with phase transitions in ecological communities

  cond-mat.stat-mech · 2025-09-29 · 3 theorem links

  A minimal neighbor-interaction Lotka-Volterra model yields exponentially many self-organized species cluster states separated by sharp phase transitions, exactly solvable via transfer matrices in the nearest-neighbor limit.

• DuoAttention: Efficient Long-Context LLM Inference with Retrieval and Streaming Heads

  cs.CL · 2024-10-14 · 3 theorem links

  DuoAttention identifies retrieval heads requiring full KV cache and streaming heads using constant-length cache to reduce memory and latency in long-context LLM inference.

• When Not to Trust Language Models: Investigating Effectiveness of Parametric and Non-Parametric Memories

  cs.CL · 2022-12-20 · 3 theorem links

  Retrieval-augmented LMs outperform much larger unassisted models on rare facts, and selective retrieval boosts performance while cutting inference costs.

• Differentiable Autoencoding Neural Operator for Interpretable and Integrable Latent Space Modeling

  cs.LG · 2025-09-30 · 3 theorem links

  DIANO builds coarse-grid latent spaces for fluid dynamics data via neural operator encoding and decoding while integrating a differentiable PDE solver directly in the latent space for end-to-end physics-constrained training.

• Markovian Reeb Graphs for Simulating Spatiotemporal Patterns of Life

  cs.CV · 2025-10-03 · 3 theorem links

  Markovian Reeb Graphs turn Reeb graphs into generative models for simulating individual and population-level spatiotemporal human mobility patterns by embedding Markovian probabilistic transitions.

• A foundation model for atomistic materials chemistry

  physics.chem-ph · 2023-12-29 · 3 theorem links

  MACE-MP-0 is a general-purpose atomistic ML force field trained on public data that enables stable simulations of diverse chemical systems with qualitative and sometimes quantitative accuracy, serving as a starting point for fine-tuning.

• Learning Linear Regression with Low-Rank Tasks in-Context

  cond-mat.dis-nn · 2025-10-06 · 3 theorem links

  The paper characterizes the distribution of predictions and generalization error for linear attention models on low-rank regression tasks in the high-dimensional limit, revealing implicit regularization from data fluctuations and a sharp phase transition governed by task structure.

• Integrability in Three-Dimensional Gravity: Eigenfunction-Forced KdV Flows

  hep-th · 2025-10-12 · 3 theorem links

  Derives forced KdV equation from Chern-Simons 3D gravity with chiral boundaries, with forcing set by Schrödinger eigenfunctions, and solves reflectionless and radiative sectors via inverse scattering.

• A Digital Twin Framework for Decision-Support and Optimization of EV Charging Infrastructure in Localized Urban Systems

  eess.SY · 2025-10-21 · 3 theorem links

  A digital twin framework integrates agent-based decision support and metaheuristic optimization to dynamically model and optimize EV charging infrastructure, policies, and renewables in a Hanoi university campus setting.

• Retrieval-Augmented Generation with Graphs (GraphRAG)

  cs.IR · 2024-12-31 · 3 theorem links

  A survey proposing a holistic GraphRAG framework with components including query processor, retriever, organizer, generator, and data source, plus domain-tailored reviews, challenges, and future directions.

• Toward Krylov-based holography in double-scaled SYK

  hep-th · 2025-10-26 · 3 theorem links

  Establishes a threefold duality linking Krylov complexity growth rate to wormhole velocity and proper momentum in DSSYK holography, with higher moments capturing replica wormholes and Krylov entropy equaling parent-geometry von Neumann entropy after tracing baby universes.

• Cambrian-S: Towards Spatial Supersensing in Video

  cs.CV · 2025-11-06 · 3 theorem links

  Cambrian-S introduces VSI-SUPER benchmarks for long-horizon spatial recall and counting, shows data scaling yields 30% gains on existing tests, and demonstrates a self-supervised next-latent predictor using surprise outperforms baselines on the new spatial supersensing tasks.

• Localization and anomalous reference frames in gravity

  hep-th · 2025-10-30 · 3 theorem links

  Constructs a phase space for gravitational degrees of freedom on null ray segments with commuting localized observables via edge modes and dressing time, then introduces an effective classical theory with Virasoro deformations to capture diffeomorphism anomalies and distinguish gauge, physical, and

• Random Walks, Faber Polynomials and Accelerated Power Methods

  math.NA · 2025-10-28 · 3 theorem links

  Random walk recurrence polynomials approximate z^n at degree ~sqrt(n) in radially convex domains, connect to Faber polynomials, and yield arbitrary-order dynamic momentum power iteration for non-symmetric matrices.

• Sweet-spot protection of hole spins in sparse arrays via spin-dependent magnetotunneling

  cond-mat.mes-hall · 2025-10-29 · 3 theorem links

  Spin-dependent magnetotunneling corrections preserve and create new sweet spots for hole spins in double quantum dots, explaining observations in shuttling and cQED experiments.

• Forget BIT, It is All about TOKEN: Towards Semantic Information Theory for LLMs

  cs.IT · 2025-11-03 · 3 theorem links

  Proposes a semantic information theory for LLMs that substitutes the token for the bit as the atomic carrier of meaning, recasts the Transformer as an energy-based model, and derives directed rate-distortion and rate-reward functions using Massey's directed information.

• Efficient Training of Language Models to Fill in the Middle

  cs.CL · 2022-07-28 · 3 theorem links

  Autoregressive language models trained on data with middle spans relocated to the end learn infilling without degrading left-to-right perplexity or sampling quality.

• Universal Differential Equations for Scientific Machine Learning

  cs.LG · 2020-01-13 · 3 theorem links

  Universal Differential Equations unify scientific models with machine learning by embedding flexible approximators into differential equations, enabling applications from biological mechanism discovery to high-dimensional optimization.

• A Survey on Knowledge Distillation of Large Language Models

  cs.CL · 2024-02-20 · 3 theorem links

  A comprehensive survey of knowledge distillation for LLMs structured around algorithms, skill enhancement, and vertical applications, highlighting data augmentation as a key enabler.

• Eagle and Finch: RWKV with Matrix-Valued States and Dynamic Recurrence

  cs.CL · 2024-04-08 · 3 theorem links

  Eagle and Finch enhance RWKV with matrix-valued states and dynamic recurrence, trained on a 1.12-trillion-token multilingual corpus, and report competitive performance on standard benchmarks.

• Smaug: Fixing Failure Modes of Preference Optimisation with DPO-Positive

  cs.CL · 2024-02-20 · 3 theorem links

  DPOP is a new loss function that prevents DPO from lowering preferred response likelihoods and outperforms standard DPO on diverse datasets, MT-Bench, and enables Smaug-72B to exceed 80% on the Open LLM Leaderboard.

• STAMP: Spatial-Temporal Adapter with Multi-Head Pooling

  cs.LG · 2025-11-13 · 3 theorem links

  STAMP adapter enables general time series foundation models to match specialized EEG foundation models on clinical classification tasks across 8 benchmarks while using few trainable parameters.

• AlphaEarth Foundations: An embedding field model for accurate and efficient global mapping from sparse label data

  cs.CV · 2025-07-29 · 3 theorem links

  AlphaEarth Foundations produces general geospatial embedding fields that consistently outperform other featurization methods on diverse mapping tasks from sparse labels and releases annual global layers for 2017-2024.

next page →

← previous page